Date of Award


Publication Type

Master Thesis

Degree Name



Electrical and Computer Engineering

First Advisor

Q. M. Jonathan Wu


Applied sciences




Stereo vision is one of the fundamental problems of computer vision. It is also one of the oldest and heavily investigated areas of 3D vision. Recent advances of stereo matching methodologies and availability of high performance and efficient algorithms along with availability of fast and affordable hardware technology, have allowed researchers to develop several stereo vision systems capable of operating at real-time. Although a multitude of such systems exist in the literature, the majority of them concentrates only on raw performance and quality rather than factors such as dimension, and power requirement, which are of significant importance in the embedded settings.

In this thesis a new miniaturized embedded stereo vision system (MESVS) is presented, which is miniaturized to fit within a package of 5x5cm, is power efficient, and cost-effective. Furthermore, through application of embedded programming techniques and careful optimization, MESVS achieves the real-time performance of 20 frames per second. This work discusses the various challenges involved regarding design and implementation of this system and the measures taken to tackle them.